The application value of deep learning-based reconstruction combined with segmented contrast injection in “one-stop” cardio-cerebrovascular computed tomography angiography

Introduction

Cardiovascular diseases are a common threat to human health worldwide, with atherosclerosis being the most common cause of cardiovascular diseases, which causes inflammatory and biochemical reactions that ultimately lead to widespread cardiovascular damage (1,2). When atherosclerosis involves a craniocervical artery, cerebral infarction can easily occur, and when it involves a coronary artery, myocardial infarction can easily result. About 20 million people die from cardiovascular and cerebrovascular diseases each year worldwide, accounting for about a third of the total global deaths (3). A close relationship between craniocervical artery disease (CCAD) and coronary artery disease (CAD) has been reported, with approximately 52% of patients with acute ischemic stroke (AIS) having symptomatic CAD and two-thirds of AIS patients with no history of heart disease having coronary artery stenosis higher than 50% (4,5). Single-site examinations often miss potential lesions in untargeted areas, so simultaneous evaluation of craniocervical and coronary arteries is of great clinical value.

Computed tomography angiography (CTA) is the main screening method for clinical cardiovascular disease evaluation due to its advantages of noninvasivity, strong operability, and ability in describing the anatomy of arteries (6,7). Traditional CTA scanning protocols are often used to image craniocervical arteries or coronary arteries alone, whereas in patients requiring multisite evaluation, repeated examination for different sites will increase the radiation dose and the risk of contrast nephropathy, as well as decrease the efficiency of the examination and diagnosis (8). In recent years, with the improvement of wide coverage detectors and scanning speed in CT technology, “one-stop” craniocervical-coronary CTA (C&C-CTA) has gradually become possible in clinical practice; however, the control of radiation dose and contrast dose has attracted further attention (9,10).

The low tube voltage technique is one of the most effective ways to reduce radiation dose and can reduce the amount of contrast by increasing the CT Hounsfield unit (HU) values of iodine, while introducing additional noise in the image. Deep learning-based reconstruction has been shown to be an appropriate method to issue this problem, and utilized in CTA for a variety of scenarios to reduce the radiation dose and amount of contrast without image quality reduction (11-14). Therefore, the aim of this study was to apply low tube voltage, deep learning image reconstruction (DLIR) algorithm in “one-stop” C&C-CTA, in order to reduce the radiation dose, contrast dose, and injection rate of patients while ensuring the image quality, and to explore the application effect of contrast segmental injection.

Methods

Participants

This prospective cohort study was approved by the Ethical Committee of Taizhou Hospital of Zhejiang Province (ethical batch No. K202306136) and conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Informed consents were provided by all patients. This study prospectively and consecutively recruited 50 adult patients suspected of having cardioembolic stroke or multi-vascular bed atherosclerosis at Taizhou Hospital of Zhejiang Province between December 2023 and April 2024. These patients underwent “one-stop” C&C-CTA (group A) examinations; simultaneously, 50 patients suspected of having CCAD were recruited for craniocervical CTA (group B), and 50 patients suspected of having CAD were recruited for coronary CTA (group C). All patients had no additional beta blockers given to control heart rate (HR) for this study. The exclusion criteria were as follows: (I) body mass index (BMI) ≤18 kg/m² or ≥25 kg/m², age <18 years old or pregnancy. (II) Poor CT image quality due to motion artifacts or metal artifacts. (III) Allergy to iodine contrast media or renal insufficiency. (IV) Anomalies of coronary artery origins or the posterior part of arterial bypass grafting (see Figure 1).

Figure 1 Flowchart of the study design. ASIR-V50%, adaptive statistical iterative reconstruction-V with 50% weighting; CT, computed tomography; CTA, computed tomography angiography; DLIR-H, deep learning image reconstruction with the high setting.

Imaging acquisition protocol

All participants scanned with a 256-detector CT scanner (Revolution Apex CT, GE HealthCare, Chicago, IL, USA). Group A was scanned with the left arm raised overhead and the right arm naturally on the side of the body, and breathing calmly. The bolus-tracking technique was employed with the region of interest (ROI) set in the descending aorta (ROI area 15–20 mm²) and triggering at the threshold of 100 HU. Craniocervical CTA was performed first with a delayed time of 5.9 seconds and ranged from the top of the skull to the aortic arch (AA), and then coronary CTA using one-beat axial scanning, was performed at an interval of another 3 seconds after the craniocervical CTA, and ranged from 2 cm below the tracheal bifurcation to the apical level. Images were acquired at 30–80% of the R-R cardiac interval. Group B was scanned with the arms naturally on the side of the body, and ranged from the AA to the top of the skull, with the trigger threshold of 90 HU, and a delay time of 3 seconds. Group C was scanned with the arms raised overhead, the trigger threshold of 150 HU, and a delay time of 3 seconds. The scan field of view (FOV) remained consistent at the same position. Detailed scan parameters are listed in Table 1. The scan protocols and contrast injection protocols for all groups are shown in Figure 2.

Figure 2 The timing and delays of image acquisition in relation to the segmented contrast media injection phases. Group A: craniocervical and coronary CTA scanning group. Group B: routine-dose groups for craniocervical CTA. Group C: routine-dose groups for coronary CTA. BMI, body mass index; CTA, computed tomography angiography.

Contrast injection protocol

Nonionic contrast agent (iohexol, 350 mgI/mL, GE HealthCare) was injected through the antecubital vein with an 18–20 G intravenous needle and a high-pressure syringe (CT Motion, Ulrich Medical, Ulm, Germany). Group A employed the segmented contrast injection protocol: administer the contrast agent at an injection rate of (0.15 × BMI) mL/s for 8 seconds, followed by the contrast agent at an injection rate of (0.18 × BMI) mL/s for 6 seconds, and then administer 50 mL of saline at an injection rate of (0.18 × BMI) mL/s. Groups B and C were injected with the contrast agent at an injection rate of (0.22 × BMI) mL/s for 10 and 8 seconds, respectively, followed by 50 mL of saline at the same injection rate.

Image post-reconstruction

Coronary imaging in both group A and group C used the smart phase technique to automatically calculate the optimal temporal phase of the cardiac cycle, and the second-generation motion correction algorithm (Snapshot Freeze 2) was used for correction of coronary motion. Images in group A were reconstructed with DLIR at high strength (DLIR-H); those in group B and group C were reconstructed with adaptive statistical iterative reconstruction-V at a strength of 50% (ASIR-V50%). All images were transferred to two types of commercial artificial intelligence software (CerebralDoc and CoronaryDoc Plus, ShuKun Technology, Beijing, China) for generation of multi planar reformation, curved planar reformation, maximal intensity projection, and volume rendering images.

Image quality evaluation

The signal-to-noise ratios (SNRs) and contrast-to-noise ratios (CNRs) were measured and calculated for each ROI in the craniocervical arteries and coronary arteries. ROIs were set at the AA, beginning of the internal carotid artery (ICA; ICA-C1), equivalent vertebral artery (VA), ICA-C4, and the M1 of the middle cerebral artery (MCA-M1) (ROI areas were 50, 10, 2, 2, and 1 mm², respectively) in craniocervical CTA image, and ROIs were set at the aortic root (AR), proximal right coronary artery (RCA), left anterior descending artery (LAD), and left circumflex artery (LCX) (ROI areas were 50, 2, 2, and 2 mm², respectively) in coronary CTA image. The CT HU values and standard deviations (SDs) of the cerebral white matter (WM) at the level of splenium of corpus callosum, the sternocleidomastoid muscle (SCM) at the level of hypopharynx and epiglottis, and the fat of the chest wall at the AR level (FAT) were measured using ROIs of 10 mm², with the SDs of the WM, SCM, and FAT taken as the background noise of the head, neck, and heart images, respectively. The ROIs were all located in the center of the target artery while avoiding arterial wall, plaque, and severe artifacts. The SNR was calculated as follows: SNR = CT_vessel/SD_vessel. Meanwhile, the CNR of the head (CNR_head), neck (CNR_neck), and heart (CNR_heart) was calculated, respectively, as follows: CNR_head = (CT_vessel − CT_WM)/SD_WM; CNR_neck = (CT_vessel − CT_SCM)/SD_SCM; CNR_heart = (CT_vessel − CT_FAT)/SD_FAT.

The sharpness of the images was assessed by the edge rise slope (ERS) and edge rise distance (ERD) of ICA-C4 (head), ICA-C1 (neck), and AR (heart) run in the axial plane (13,15). Profile curves were generated using Image J software (National Institutes of Health, Bethesda, MD, USA) (http://rsb.info.nih.gov/ij) and its particle analysis tool (Plot Profile). The start and the end points of the CT HU values curve were set to the center of the vessel and a low CT HU value outside the vessel, respectively. We measured 10–90% of the ERD as the edge response width at the boundary of the vascular and calculated the ERS [ERS = (CT90% − CT10%)/ERD] (16) (Figure S1). The higher the ERS, the sharper was the edge.

Two experienced radiologists (with 10 and 5 years of experience in CTA images, respectively) blindly evaluated the subjective quality of images. A five-point Likert scale was used to grade the artery enhancement, clarity of small arterial details, image noise, and artifact (scores greater than 3 were regarded as the image being diagnostic): 5= excellent artery enhancement, clear anatomical details, subtle image noise and no artifacts; 4= good artery enhancement, relatively clear anatomical structures and details, subtle image noise and slight artifacts; 3= acceptable artery enhancement, majority of clear anatomical structures, moderate image noise and some artifacts; 2= poor artery enhancement, limited delineation of tissue structural details, substantial image noise and significant artifacts; and 1= poor artery enhancement, unclear anatomical structures, severe image noise and serious artifacts. In case of disagreement, the final score in was decided by mutual agreement.

Radiation dose

Volume CT dose index (CTDIvol; mGy) and dose-length product (DLP; mGy·cm), recorded in the patient protocols, were used to calculate the radiation dose. Then, the effective dose (ED; mSv) was defined as the product of the DLP and a conversion factor k (for craniocervical imaging, k =0.0031 mSv·mGy⁻¹·cm⁻¹; for coronary imaging, k =0.014 mSv·mGy⁻¹·cm⁻¹) (17).

Statistical analysis

All statistical analyses in this study were performed with the SPSS version 26.0 statistical software (IBM Corp., Armonk, NY, USA). Continuous variables were presented as mean ± SD and tested for normality using the Shapiro-Wilk test. Non-parametric variables were expressed as the median with interquartile range and presented in the form of frequency distribution tables. Normally distributed variables were compared using one-way analysis of variance (ANOVA) with least significant difference (LSD) post-hoc correction (multiple groups) and independent Student t-tests (two groups). The non-parametric variables were analyzed using the Friedman test with Nemenyi post-hoc correction (multiple groups) and Mann-Whitney U test (two groups). The Fisher exact test was used to compare categorical variables. The kappa statistic was used to evaluate the interrater agreement (κ value >0.8 was defined as excellent, 0.6≤ κ value ≤0.8 was defined as good, and κ value <0.6 was defined as bad agreement). A P value <0.05 was considered statistically significant.

Results

Participants

A total of 150 patients were enrolled, 80 of whom were male and 70 female, with age of 57.10±10.85 years and BMI of 22.23±1.35 kg/m². As shown in Table 2, the age, gender and BMI were comparable between groups (age, P=0.14; gender, P=0.83; BMI, P=0.34). The HR was not statistically significantly different between groups A and C (P=0.20), in which there were 29 cases of arrhythmia or HR >90 bpm.

There were significant differences in contrast dose, injection rate, and ED among three groups (all P<0.001). The contrast dose in group A (51.44±3.89 mL) was higher than those in groups B (48.32±2.93 mL) and C (39.24±2.11 mL), but decreased by 41.25% compared with the sum of groups B and C. The injection rate in group A (3.67±0.28 mL/s) was lower than those in groups B (4.83±0.29 mL/s) and C (4.88±0.2 mL/s). The ED in group A (3.39±0.43 mSv) was higher than that in group B (1.37±0.13 mSv), lower than group C (4.65±0.75 mSv), and 43.78% lower than the sum of groups B and C (see Table 2).

Objective evaluation

Craniocervical artery

The CT HU values of group A were higher than group B and were statistically different (all P<0.001). In the head, background noise in group A was lower than group B and statistically different (12.82±2.52 vs. 14.93±2.00 HU, P<0.001); meanwhile, in the neck, the background noise was not statistically different (7.66±1.98 vs. 8.14±2.15 HU, P=0.25). SNRs were higher in group A for all vessels and were statistically different from those in group B (all P<0.05) except VA and ICA-C4 (SNR_{A, VA} vs. SNR_{B, VA}, P=0.53; SNR_{A, ICA-C4} vs. SNR_{B, ICA-C4}, P=0.33). CNRs were higher in group A, with all vessels showing statistically significant differences from group B (all P<0.001).

ERSs were higher and ERDs were lower in group A than they were in group B, and both were statistically different (all P<0.05) (Table 3 and Figure 3).

Figure 3 Comparison of the objective evaluation of each vessel. Group A: craniocervical and coronary CTA scanning group. Group B: routine-dose groups for craniocervical CTA. Group C: routine-dose groups for coronary CTA. AA, aortic arch; AR, aortic root; CTA, computed tomography angiography; CNR, contrast-to-noise ratio; CT, computed tomography; ERS, edge rise slope; ERD, edge rise distance; FAT, aortic root-level chest wall fat; ICA-C1, internal carotid artery beginning segment; ICA-C4, internal carotid artery C4 segment; LAD, left anterior descending artery; LCX, left circumflex artery; MCA-M1, middle cerebral artery M1 segment; RCA, right coronary artery; SD, standard deviation; SCM, sternocleidomastoid muscle; SNR, signal-to-noise ratio; VA, vertebral artery; WM, white matter; HU, hounsfield unit.

Coronary artery

The CT HU values of AR in group A were lower than those in group C and statistically different (P=0.02), but not statistically different from other vessels (all P>0.05). The background noise in group A was lower than that in group C and statistically different (12.84±2.66 vs. 18.21±4.21 HU, P<0.001). The SNR of RCA was not statistically significantly different between two groups (P=0.92), and the SNR of all other vessels was higher in group A than it was in group C and both were statistically different (all P<0.05). CNRs were higher in group A, with all vessels showing statistically significant differences from group C (all P<0.001).

ERS and ERD in group A were not statistically different from those in group C (ERS_{A, AR} vs. ERS_{C, AR}, P=0.11; ERD_{A, AR} vs. ERD_{C, AR}, P=0.18) (Table 3 and Figure 3).

Subjective evaluation

All scores were higher than 3, meeting the diagnostic requirement. Subjective evaluation showed no statistically significant differences in either craniocervical or coronary arteries (P>0.05). The κ values were all higher than 0.88, indicating excellent consistency between the two readers (Table 4 and Figures 4,5).

Figure 4 Image quality comparison between the two groups. (A-E) The reconstructed images of the craniocervical CTA in group A, and (F-J) the reconstructed images from group B. There was no statistically significant difference on subjective quality score between the two groups. Group A: craniocervical and coronary CTA scanning group. Group B: routine-dose groups for craniocervical CTA. CTA, computed tomography angiography.

Figure 5 Image quality comparison between the two groups. (A-D) The reconstructed images of the coronary CTA in group A, and (E-H) the reconstructed images from group C. There was no statistically significant difference on subjective quality score between the two groups. Group A: craniocervical and coronary CTA scanning group. Group C: routine-dose groups for coronary CTA. CTA, computed tomography angiography.

Discussion

The results demonstrated that for patients requiring simultaneous examination of the craniocervical and coronary arteries, using a low tube voltage, combined with the DLIR-H algorithm and a contrast segmental injection protocol, could reduce ED, contrast dose, and injection rate by 43.78%, 41.25%, and 24.49%, respectively, without compromising image quality.

Atherosclerosis is a chronic, progressive polyvascular disease that primarily affects large- and medium-sized arteries in the body. Both the craniocarotid and coronary arteries are medium-sized musculoelastic arteries of the circulatory system and are often involved simultaneously due to their shared pathophysiology characteristics, mechanisms, and risk factors (18). Studies have shown that patients with concurrent craniocervical and coronary artery atherosclerosis are at a higher risks of major adverse vascular events compared to those with isolated craniocervical or coronary artery atherosclerosis. Additionally, craniocervical atherosclerosis can serve as an independent predictor of major adverse cardiac events. Therefore, systematic evaluation of arterial diseases in high-risk populations is crucial for guiding clinical strategies, reducing the likelihood of missed diagnosis, and improving patient outcomes (19,20). In this study, 18 patients (36%) in group A were diagnosed with both CAD and CCAD. The “one-stop” C&C-CTA protocol offers a safer and more efficient screening approach by minimizing the burden of pre-scan preparations, reducing time costs, and enabling earlier clinical interventions. However, due to the higher radiation and contrast doses, as well as lack of relevant studies and established guidelines, the “one-stop” C&C-CTA remains impractical as first line examination for patients suspected of having either CAD or CCAD alone.

Various strategies for reducing radiation dose have been implemented in CT examinations, including low tube voltage, automatic tube current modulation, and iterative reconstruction, with low tube voltage technology being the primary technique for radiation dose reduction (21,22). Lower energy photons produce more attenuation, thus vessel visualization is improved, allowing for a reduction in contrast agent dose. However, combining low tube voltage with reduced contrast dose increases image noise and degrades the image quality. Although the noise reduction capability of iterative reconstruction has been demonstrated in some previous studies, significant reductions in radiation dose can cause the algorithm to damage the image texture in an attempt to enhance the overall image quality, limiting its potential for dose reduction (23,24). In this study, DLIR-H was employed to compare with ASIR-V50%, and the results showed that the DLIR-H enhanced the sharpness of craniocervical vessels and reduced the background noise in head and heart regions.

In a previous study on “one-stop” C&C-CTA, Li et al. (25) found that reducing the tube voltage from 100 to 80 kVp decreased the contrast dose by 30% (51.13±6.91 to 35.80±4.85 mL) and the radiation dose by 48% (1.91±0.42 to 1.00±0.09 mSv). Similarly, Zhao et al. (9) reported a 51% reduction in contrast dose (60 vs. 40 mL) and a 59% reduction in radiation dose (0.79±0.08 vs. 0.32±0.11 mSv) when the tube voltage was decreased from 100 kVp to either 70 kVp or 80 kVp in accordance with body weight. In our study, reducing the tube voltage from 100 to 80 kVp led to a 41.25% reduction in contrast dose (87.56±3.51 to 51.44±3.89 mL) and a 43.78% reduction in radiation dose (6.03±0.72 to 3.39±0.43 mSv). Two key factors explain the differences in radiation dose reduction between our study and others. First, to minimize the increasing of noise from the arms and to compensate for the reducing of X-ray energy at lower tube voltages, we used a much smaller noise index (NI) (NI for craniocervical CTA =6, NI for coronary CTA =15) and a wider mA scanning range (100–1,080 mA) compared to other studies. Second, to capture dynamic changes in cardiac structure and function throughout the cardiac cycle and provide a more comprehensive view for the diagnosing and treating cardiovascular diseases, we employed a larger cardiac cycle (30–80%) scanning mode (10). These factors likely contributed to the higher radiation dose observed in our study. In “one-stop” C&C-CTA scanning, the CT HU value at the first scanning site is typically higher than that at the second scanning site. When the CT HU value at the first scanning site is maintained within the standard diagnostic range, the CT HU value at the second scanning site tends to drop, potentially failing to meet the diagnostic requirements. Conversely, ensuring the CT HU value at the second scanning site is within the standard diagnostic range can cause the CT HU value at the first scanning site increase cumulatively due to the extended injection of the contrast agent. This continuous enhancement can interfere with the detection of microplaques (26,27). To mitigate this issue, we applied a segmented contrast injection protocol by lowering the injection rate during the first 8 seconds (from BMI × 0.18 to BMI × 0.15). The results indicated that the CT HU values across all vascular segments met diagnostic criteria. Despite a 30.5% reduction in the initial injection rate in group A compared to group B, the CT HU values in the craniocervical region remained significantly higher than those in group B and statistically different (P<0.05). This difference can be attributed to both the cumulative effect of long injection duration and the lower tube voltage used in group A, which aligns the average energy of the X-rays more closely with the k-edge of iodine, thereby enhancing iodine attenuation in the target vessels. These findings suggest that the first segment injection rate could be further reduced. In summary, the segmented contrast injection protocol can be successfully incorporated into “one-stop” C&C-CTA protocols, which help to reduce the injection rate and contrast dose, as well as decrease the rate of extravasation of contrast and the risk of contrast nephropathy.

There were several limitations in this study: (I) this study focused on the Image quality closely related to diagnosis to evaluate the feasibility of “one-stop” C&C-CTA protocols; a comparison with accuracy of imaging diagnostic efficacy remains to be explored. (II) This study analyzed only patients with normal BMI range, limiting its applicability in obese patients. (III) The design of the segmented contrast injection protocol was subjective, and will be studied further to optimization. (IV) This study presents a parameter optimization strategy tailored for specific CT scanning systems and their supporting software platforms, which can also be applied to similar systems (e.g., low-dose CT platforms with deep learning reconstruction techniques). However, the actual application necessitates validation and calibration in accordance with the specific characteristics of each device.

Conclusions

Compared to scanning the coronary and craniocervical artery individually, cardio-cerebrovascular CTA based on DLIR can reduce the radiation dose, contrast dose. and injection rate without compromising image quality.

Acknowledgments

None.

Footnote

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2024-2579/dss

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2024-2579/coif). A.S. is an employee of GE Healthcare China. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This prospective cohort study was approved by the Ethical Committee of Taizhou Hospital of Zhejiang Province (ethical batch No. K202306136). Informed consents were obtained from all patients.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Joint Committee on the Chinese Guidelines for Lipid Management. Chinese guidelines for lipid management (2023). Zhonghua Xin Xue Guan Bing Za Zhi 2023;51:221-55. [Crossref] [PubMed]
2. Zhou M, Wang H, Zeng X, Yin P, Zhu J, Chen W, et al. Mortality, morbidity, and risk factors in China and its provinces, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 2019;394:1145-58. [Crossref] [PubMed]
3. World Heart Federation. World Heart Report 2023: Confronting the World’s Number One Killer. 2023. Available online: https://world-heart-federation.org/resource/world-heart-report-2023/#
4. Gunnoo T, Hasan N, Khan MS, Slark J, Bentley P, Sharma P. Quantifying the risk of heart disease following acute ischaemic stroke: a meta-analysis of over 50,000 participants. BMJ Open 2016;6:e009535. [Crossref] [PubMed]
5. Yoo J, Song D, Baek JH, Kim K, Kim J, Song TJ, Lee HS, Choi D, Kim YD, Nam HS, Heo JH. Poor long-term outcomes in stroke patients with asymptomatic coronary artery disease in heart CT. Atherosclerosis 2017;265:7-13. [Crossref] [PubMed]
6. Bao LW, Wang ZF, Sun ZT, Meng LX, Zhao L. Relationship of atherosclerosis between carotid arteries and coronary arteries by CTA evaluation: A retrospective study. CT Theory and Applications 2017;26:497-504.
7. Caetano Dos Santos FL, Kolasa M, Terada M, Salenius J, Eskola H, Paci M. VASIM: an automated tool for the quantification of carotid atherosclerosis by computed tomography angiography. Int J Cardiovasc Imaging 2019;35:1149-59. [Crossref] [PubMed]
8. Davenport MS, Perazella MA, Yee J, Dillman JR, Fine D, McDonald RJ, Rodby RA, Wang CL, Weinreb JC. Use of Intravenous Iodinated Contrast Media in Patients With Kidney Disease: Consensus Statements from the American College of Radiology and the National Kidney Foundation. Kidney Med 2020;2:85-93. [Crossref] [PubMed]
9. Zhao L, Bao J, Guo Y, Li J, Yang X, Lv T, Hao F, Wang Z, Yang Z, Liu A. Ultra-low dose one-step CT angiography for coronary, carotid and cerebral arteries using 128-slice dual-source CT: A feasibility study. Exp Ther Med 2019;17:4167-75. [Crossref] [PubMed]
10. Lee S, Kim IC, Kim YD, Nam HS, Kim SY, Choi SM, Chang HJ. The role of cardiac CT throughout the full cardiac cycle in diagnosing patent foramen ovale in patients with acute stroke. Eur Radiol 2021;31:8983-90. [Crossref] [PubMed]
11. Hsieh J, Liu E, Nett B, Tang J, Thibault JB, Sahney S. A new era of image reconstruction: TrueFidelity™ Technical white paper on deep learning image reconstruction. 2019.
12. Li W, Diao K, Wen Y, Shuai T, You Y, Zhao J, Liao K, Lu C, Yu J, He Y, Li Z. High-strength deep learning image reconstruction in coronary CT angiography at 70-kVp tube voltage significantly improves image quality and reduces both radiation and contrast doses. Eur Radiol 2022;32:2912-20. [Crossref] [PubMed]
13. Tatsugami F, Higaki T, Nakamura Y, Yu Z, Zhou J, Lu Y, Fujioka C, Kitagawa T, Kihara Y, Iida M, Awai K. Deep learning-based image restoration algorithm for coronary CT angiography. Eur Radiol 2019;29:5322-9. [Crossref] [PubMed]
14. Zhang Q, Lin Y, Zhang H, Ding J, Pan J, Zhang S. The application value of a vendor-specific deep learning image reconstruction algorithm in "triple low" head and neck computed tomography angiography. Quant Imaging Med Surg 2024;14:2955-67. [Crossref] [PubMed]
15. Cao L, Liu X, Qu T, Cheng Y, Li J, Li Y, Chen L, Niu X, Tian Q, Guo J. Improving spatial resolution and diagnostic confidence with thinner slice and deep learning image reconstruction in contrast-enhanced abdominal CT. Eur Radiol 2023;33:1603-11. [Crossref] [PubMed]
16. Tatsugami F, Higaki T, Sakane H, Fukumoto W, Kaichi Y, Iida M, Baba Y, Kiguchi M, Kihara Y, Tsushima S, Awai K. Coronary Artery Stent Evaluation with Model-based Iterative Reconstruction at Coronary CT Angiography. Acad Radiol 2017;24:975-81. [Crossref] [PubMed]
17. AAPM Report No. 96. The Measurement, Reporting, and Management of Radiation Dose in CT. Report of AAPM TaskGroup 23 of the Diagnostic Imaging Council CT Committee. American Association of Physicists in Medicine; 2008.
18. Sigala F, Oikonomou E, Antonopoulos AS, Galyfos G, Tousoulis D. Coronary versus carotid artery plaques. Similarities and differences regarding biomarkers morphology and prognosis. Curr Opin Pharmacol 2018;39:9-18. [Crossref] [PubMed]
19. Roh JW, Kwon BJ, Ihm SH, Lim S, Park CS, Chang K, Chung WS, Kim DB, Kim SR, Kim HY. Predictors of Significant Coronary Artery Disease in Patients with Cerebral Artery Atherosclerosis. Cerebrovasc Dis 2019;48:226-35. [Crossref] [PubMed]
20. Guo R, Deng J, Rong P, Zhou W, Zhang G, Peng S, Liang Q, Yang X, Hu P. One-stop combined CT angiography of coronary and craniocervical arteries: recommended as the first examination for patients suspected of coronary or craniocervical artery disease. Eur Radiol 2023;33:7034-43. [Crossref] [PubMed]
21. Lell MM, Kachelrieß M. Recent and Upcoming Technological Developments in Computed Tomography: High Speed, Low Dose, Deep Learning, Multienergy. Invest Radiol 2020;55:8-19. [Crossref] [PubMed]
22. Jia CF, Zhong J, Meng XY, Sun XX, Yang ZQ, Zou YJ, Wang XY, Pan S, Yin D, Wang ZQ. Image quality and diagnostic value of ultra low-voltage, ultra low-contrast coronary CT angiography. Eur Radiol 2019;29:3678-85. [Crossref] [PubMed]
23. Stocker TJ, Leipsic J, Hadamitzky M, Chen MY, Rubinshtein R, Deseive S, Heckner M, Bax JJ, Kitagawa K, Marques H, Schmermund A, Silva C, Mahmarian J, Kang JW, Grove EL, Lesser J, Massberg S, Hausleiter J. Application of Low Tube Potentials in CCTA: Results From the PROTECTION VI Study. JACC Cardiovasc Imaging 2020;13:425-34. [Crossref] [PubMed]
24. Pauchard B, Higashigaito K, Lamri-Senouci A, Knebel JF, Berthold D, Verdun FR, Alkadhi H, Schmidt S. Iterative Reconstructions in Reduced-Dose CT: Which Type Ensures Diagnostic Image Quality in Young Oncology Patients? Acad Radiol 2017;24:1114-24. [Crossref] [PubMed]
25. Li W, Huang W, Li P, Wen Y, Shuai T, He Y, You Y, Yu J, Diao K, Song B. Application of deep learning image reconstruction-high algorithm in one-stop coronary and carotid-cerebrovascular CT angiography with low radiation and contrast doses. Quant Imaging Med Surg 2024;14:1860-72. [Crossref] [PubMed]
26. Tognolini A, Arellano CS, Marfori W, Heidari G, Sayre JW, Krishnam MS, Ruehm SG. Comprehensive low-dose imaging of carotid and coronary arteries with a single-injection dual-source CT angiography protocol. Clin Radiol 2014;69:246-53. [Crossref] [PubMed]
27. Takagi H, Leipsic JA, Indraratna P, Gulsin G, Khasanova E, Tzimas G, et al. Association of Tube Voltage With Plaque Composition on Coronary CT Angiography: Results From PARADIGM Registry. JACC Cardiovasc Imaging 2021;14:2429-40. [Crossref] [PubMed]